AI Meets Agriculture Building Food Security and Climate Resilien

20 Feb 2026 10:00h - 11:00h

Session at a glance

Summary

This discussion focused on using artificial intelligence to enhance food security and climate resilience in agriculture, with particular emphasis on Maharashtra’s pioneering AI initiatives and the broader implications for India and developing nations. The session was part of the India AI Impact Summit and featured key government officials, international development experts, and technology leaders discussing the implementation of AI-driven agricultural systems at scale.

Chief Minister Devendra Fadnavis outlined Maharashtra’s comprehensive Maha Agri AI Policy 2025-2029, highlighting the state’s transition from pilot projects to full-scale implementation through platforms like Mahavistar, which serves over 2.5 million farmers with multilingual advisories. Dr. Devesh Chaturvedi from the Ministry of Agriculture emphasized the need to move beyond “digital red tapism” by integrating various agricultural services into unified AI platforms, supported by the development of 9 crore farmer IDs as part of India’s digital public infrastructure.

The World Bank’s Johannes Zutt stressed the importance of collaborative ecosystems where government provides foundational infrastructure and governance, while private sector innovation drives creative applications. He highlighted the potential for India’s diverse agricultural challenges to generate solutions applicable globally through South-South knowledge exchange.

Dr. Soumya Swaminathan raised critical concerns about ensuring AI systems include women farmers, who are increasingly central to agriculture but often lack formal land ownership and may be excluded from data systems. She emphasized the need for human oversight, iterative evaluation, and design principles that reduce drudgery while maintaining scientific integrity.

Shankar Maruwada from Ekstep Foundation drew parallels between historical agricultural revolutions and the current AI transformation, advocating for open, interoperable systems based on DPI principles. He emphasized that collaborative, network-based approaches rather than siloed platforms would enable rapid diffusion of innovations across states and sectors.

The discussion concluded with plans for the AI4Agri 2026 Global Conference in Mumbai, positioning India as a leader in responsible AI deployment for agriculture at population scale.

Keypoints

Major Discussion Points:

– Maharashtra’s AI Agriculture Policy and Implementation: The state has launched the Maha Agri AI Policy 2025-2029, featuring platforms like Mahavistar (serving 2.5+ million farmers with multilingual advisories), Agristrack (for seamless access to schemes), and Maha AgEx (open data exchange architecture). This represents a shift from pilot projects to full-scale AI deployment in agriculture.

– Digital Public Infrastructure (DPI) and Interoperability: Discussion centered on building open, interoperable systems similar to India’s successful DPI model (like UPI), with emphasis on creating “shared rails” that allow different stakeholders to plug in services while maintaining common standards and avoiding fragmented, siloed systems.

– Inclusion and Gender Equity in AI Agriculture: Significant focus on ensuring women farmers are not left behind in AI transformation, addressing challenges like lack of land ownership documentation, different farming practices, and the need for AI systems designed with women’s specific needs in mind, especially as 2026 is the International Year of Women in Agriculture.

– Central-State Collaboration Framework: Discussion of how to balance national AI architecture (like Bharat Vistar) with state-level innovation flexibility, emphasizing the need for coordinated efforts between central government, states, and various stakeholders while maintaining interoperability and data trust.

– Responsible AI Governance and Trust: Emphasis on building AI systems with transparency, accountability, scientific integrity, and farmer consent, moving beyond “digital red tapism” to create trustworthy, explainable AI that serves farmers rather than exploiting them.

Overall Purpose:

The discussion aimed to move from vision to implementation for AI in agriculture, specifically focusing on how to institutionalize AI within agricultural systems at population scale while ensuring inclusion, interoperability, and sustainable governance. It served as a precursor to the AI4Agri 2026 Global Conference, bringing together policy leaders, development experts, and technology innovators to establish frameworks for responsible AI deployment in agriculture.

Overall Tone:

The discussion maintained an optimistic and collaborative tone throughout, characterized by visionary leadership and practical problem-solving. Speakers demonstrated enthusiasm for AI’s transformative potential while maintaining a responsible, cautious approach to implementation. The tone was consistently forward-looking and inclusive, with participants building on each other’s ideas rather than presenting conflicting viewpoints. There was a strong sense of urgency balanced with careful consideration of ethical implications and the need for inclusive design principles.

Speakers

Speakers from the provided list:

– Vikas Chandra Rastogi – Secretary, Ministry of Agriculture and Farmers’ Welfare, Government of Maharashtra (moderator/host of the session)

– Dr. Soumya Swaminathan – Chairperson of Dr. M.S. Swaminathan Research Foundation; global leader in science, champion for sustainable development, and advocate for mainstreaming women farmers’ roles in agriculture

– Devendra Fadnavis – Honorable Chief Minister of Maharashtra

– Shankar Maruwada – Co-founder and CEO of Ekstey Foundation; pioneer in building digital public infrastructure

– Johannes Zutt – Regional Vice President, World Bank

– Devesh Chaturvedi – Secretary, Ministry of Agriculture and Farmer Welfare, Government of India

Additional speakers:

– Ashish Shailar – Honorable Minister (specific portfolio not mentioned)

– Nitesh Rane – Minister/Government official (specific role not detailed)

Full session report

This discussion at the India AI Impact Summit focused on leveraging artificial intelligence to enhance food security and climate resilience in agriculture, with Maharashtra’s initiatives serving as a model for national and global implementation. The session brought together government officials, international development experts, and technology leaders to address scaling AI from pilot projects to population-level deployment.

Opening Context and Platform Overview

Vikas Chandra Rastogi, moderating the session, outlined the current landscape of AI agriculture platforms. He highlighted Maharashtra’s three key initiatives: Mahavistar, an AI-powered mobile platform serving over 2.5 million farmers with multilingual advisories, market intelligence, and pest alerts; Agristrack for government scheme access; and Maha AgEx as a federated data exchange architecture. Significantly, Mahavistar has integrated Bili, the first tribal language in India, demonstrating commitment to inclusive design.

Maharashtra’s AI Agriculture Policy Framework

Chief Minister Devendra Fadnavis presented Maharashtra’s Maha Agri AI Policy 2025-2029, emphasizing the shift from demonstration projects to full-scale implementation. “We are moving from pilots to policy, from experiments to ecosystem, and from fragmented approaches to federated architecture,” he stated. The policy is built on principles of openness and interoperability, creating scalable public infrastructure rather than isolated solutions.

The Chief Minister outlined Maharashtra’s advantages for agri-innovation, including 150 lakh hectares of cultivated land, diverse agro-climatic conditions, leading agricultural universities, and a vibrant startup ecosystem. He emphasized that AI must be built on “trusted data, ethical governance, and public accountability,” noting that without trust, scale cannot be achieved.

Maharashtra has implemented AI-based pest surveillance integrated with geospatial analytics, delivering early warnings to cotton farmers and reducing crop vulnerability. The state is developing a statewide interoperable agriculture data exchange based on open standards and robust data governance to ensure data empowers rather than exploits farmers.

For the upcoming AI for Agri 2026 Global Conference (February 22-23 at Jio World Convention Center, Mumbai), the Chief Minister outlined four foundational pillars: responsible governance, open and interoperable digital infrastructure, investment and scaling, and inclusion and gender equity.

National Framework and Digital Public Infrastructure

Dr. Devesh Chaturvedi, Secretary of the Ministry of Agriculture and Farmers’ Welfare, addressed the challenge of “digital red tapism” created by multiple separate applications for different agricultural schemes. “While we successfully reduced bureaucratic barriers through initial digitization, we inadvertently created new complexities,” he explained.

The ministry’s response has been Bharatvistar, India’s first integrated AI-based system for farmers, providing weather advisories, ICR-based crop advisories, pest information, and government scheme details through both mobile applications and telephony. Currently operating in English and Hindi, the system will expand to all Bhashini-related languages within three to six months.

Central to this framework is the creation of “close to 9 crore farmer IDs” containing comprehensive information about crops, land holdings, soil health, and cultivation patterns. Dr. Chaturvedi explained that integration of farmer IDs with AI advisory systems, planned within three to six months, will enable consent-based access to farmer data for highly personalized guidance.

He also highlighted a predictive model serving 3.8 crore farmers using 100 years of IMD data for monsoon prediction, demonstrating the scale of AI implementation already underway.

Global Development Perspective

Johannes Zutt from the World Bank, drawing from his personal farming background, outlined distinct stakeholder roles in AI agriculture ecosystems. Government responsibilities include establishing foundational infrastructure, ensuring AI governance and interoperability, developing educational programs addressing digital literacy, and maintaining scientific credibility of research and extension services.

The private sector contributes through creative application development, with Zutt citing a Moroccan example where a tomato farmer developed an application determining precise water requirements by photographing plants. International institutions can provide financing and advisory support while ensuring information credibility.

Zutt emphasized India’s unique position to lead AI development for developing countries: “India’s enormous population diversity, multiple languages, varied regional conditions, and different cultural contexts create a natural laboratory for solutions applicable globally.”

Inclusion and Gender Equity Challenges

Dr. Soumya Swaminathan highlighted critical inclusion challenges, particularly for women farmers who increasingly assume farming responsibilities as men migrate to cities. She noted that only approximately 25% of agricultural properties are registered in women’s names, creating systematic exclusion risks from AI systems that rely on publicly available data.

“Algorithms trained on available data may develop recommendations suitable for male farmers operating tractors but irrelevant for women using traditional tools and methods,” she warned. Dr. Swaminathan advocated for rigorous evaluation processes similar to medical research, including systematic bias assessment and identification of excluded populations.

She emphasized examples of inclusive technology design, mentioning the “Fisher Friendly Mobile App” and “Women Connect app” as models. Crucially, she argued for maintaining “humans in the loop” rather than pursuing complete automation, both for quality control and employment considerations, suggesting AI should augment rather than replace human agricultural extension workers.

Technical Architecture and Implementation Philosophy

Shankar Maruwada from Ekstep Foundation drew historical parallels between current AI transformation and the Haber-Bosch process, noting that breakthrough technologies require comprehensive ecosystem development. He described the current AI revolution as “pulling intelligence from the earth” and providing it to farmers.

Mahavistar’s design exemplified inclusive architecture by prioritizing service to illiterate farmers using basic feature phones, enabling natural language conversations in local dialects. The technical architecture follows Digital Public Infrastructure principles, creating “shared rails” similar to India’s railway network—common infrastructure enabling diverse applications while maintaining interoperability.

Maruwada emphasized that AI systems improve through usage rather than requiring perfection before deployment, advocating a “deploy and iterate” approach. He highlighted collaborative development involving the IndiAI Mission, Bhashini, IIT Madras, IIIT Hyderabad, and other partners, demonstrating the networked approach to building these systems.

Implementation Challenges and Governance

Throughout the discussion, speakers emphasized that successful AI agriculture deployment requires robust governance frameworks ensuring transparency, accountability, and farmer trust. Key challenges include ensuring scientific credibility of AI-generated advice, managing transitions from fragmented databases to integrated systems, and balancing innovation with safety requirements.

The consensus favored gradual, human-supervised implementation rather than complete automation, addressing both quality control concerns and employment implications while enabling continuous improvement based on farmer feedback.

Investment and Collaboration Framework

The Chief Minister positioned Maharashtra as offering a compelling agri-innovation ecosystem, actively inviting venture capital funds, impact investors, multilateral development banks, and philanthropic foundations to partner in scaling AI solutions. This framework emphasizes that partnerships with Maharashtra enable development of scalable solutions applicable to emerging economies worldwide.

The session demonstrated significant progress in AI agriculture transformation while acknowledging ongoing challenges in inclusion, connectivity, and maintaining scientific credibility. The combination of political leadership, technical innovation, collaborative partnerships, and commitment to inclusive design creates opportunities for AI to enhance food security and climate resilience across India and beyond.

Session transcript

Vikas Chandra Rastogi

May I invite Dr. Devish Chaturvedi, Secretary, Ministry of Agriculture and Farmers’ Welfare. Sir, please come onto the stage. Sir, please come onto the stage. Johannes Jett, Regional Vice President, World Bank. stage please. Honourable Chief Minister of Maharashtra, Shri Devendra Farnavis Ji. Honourable Minister, Shri Ashish Shailar Ji, Shri Nitesh Rane Ji. Our distinguished guests from India and around the world, very good morning. On behalf of the Government of Maharashtra, I welcome you to the session on Using AI for Food and Climate Resilience. Agriculture is at a turning point. Climate change is making farming riskier, resources are limited and markets are changing quickly. However, there is an opportunity. Digital tools and AI are advancing fast. Our goal is not just to use AI tools.

We must build intelligence into our public systems to help everyone. For India, the change is essential. It is the key to food and nutrition security, higher farmer incomes, and a stable economy. India has shown that digital systems work when they are open and well -governed. Our next step is to bring AI into this framework in a responsible way. Under the leadership of Honourable Chief Minister of Maharashtra, the state has launched the Maha Agri AI Policy 2025 -2029. This policy uses AI for farmer advisory services, market information, data exchange, product traceability, innovation and research, and creating capacities of stakeholders. Thank you. We are moving beyond pilots to project… at full scale. Mahavistar is the country’s first AI -powered network and information and advisory services.

Today, Mahavistar is being used by more than 2 .5 million farmers to get advisories in Marathi language and recently, the first tribal language in the country, Bili, has also been integrated into Mahavistar. Agristrack is helping farmers to get seamless access to various schemes and services. The Maha AgEx, which is an open, federated and consent -driven architecture for data exchange, it is helping us to bring diverse data sets together to get us a big picture. Agriculture is now a key part of India’s AI mission. We are proud to work with the Government of India to lead this change. I want to thank the Ministry of Electronics and Information Technology, Ministry of Agriculture, Extra Foundation, and the Department of Agriculture, and the Department of Agriculture, and the Department of Agriculture, and the Department of Agriculture, and the Department of Agriculture, and the Department of Agriculture, and the Department of Agriculture, and the Department of Agriculture, the World Bank, MS Swaminathan Research Foundation, the Gates Foundation, and all our partners for their support.

It is now my duty to invite our Honorable Chief Minister to the stage. He will share his vision for using AI to strengthen our food systems and protect our climate. After the address of Honorable Chief Minister, we have a panel discussion with our distinguished panelists. Welcome.

Devendra Fadnavis

A very good morning to all of you. Shri Devesh Chaturvedi, Rajesh Agarwal, Vikas Rastogi, Mr. Jonas Jett, Shubhati Swaminathan, Shushankar Maruwada, my colleagues, Shashi Shailarji, Nitesh Raneji, all the dignitaries present here. Namaskar and good morning to everyone. It is my privilege to address this distinguished gathering at the India AI Impact Summit and this important session on AI in Agriculture. We meet at a very defining moment across the world. Food systems are under strain. Climate volatility is intensifying. Water tables are falling. Soil health is deteriorating. Supply chains are fragile and global markets are unpredictable. For countries from the global south, agriculture is not merely an economic challenge. sector. It is livelihood, social stability, and national security.

India understands this very deeply. And under the visionary leadership of our Honorable Prime Minister Narendra Modi, India has placed digital public infrastructure and responsible AI at the center stage of national development. The India AI mission is about using technology to deliver inclusion, transparency, and scale. Today, agriculture must sit at the heart of this mission. Over half a billion Indians depend directly or indirectly on agriculture. Yet, smallholders face fragmented information, rising input costs, climate uncertainty, and limited access to credit and market. Traditional extension systems, however committed, cannot match the scale and the speed required. Artificial intelligence changes this equation. AI can provide hyperlocal weather predictions, early pest outbreaks, warnings, precision irrigation and fertilizer guidance, credit scoring based on crop intelligence, transparent traceable supply chains, real -time market advisories.

But let me emphasize, AI is not a magic. As Honorable PM said in his inaugural session, AI must be built on trusted data, ethical governance. And public accountability. Without trust, scale will not happen. Last year, Maharashtra made a very clear and decisive strategic decision AI in agriculture must not remain confined to demonstrations or pilots It must reach millions Under our Maha Agri AI policy 2025 -29 We adopted a policy -led ecosystem -driven model Built on openness and interoperability Allow me to share what this has meant in practice As rightly told by our Secretary Maha Vistar Our AI -powered mobile platform delivers multilingual personalized advisories Market intelligence, pest alerts and access to government services More than 2 .5 million downloads Acting as a platform for AI -powered mobile platform The Maha Agri AI is a platform for AI -powered mobile platform The Maha Agri AI is a platform for AI -powered mobile platform digital friend to all these farmers.

This demonstrates one thing very clearly. Farmers are ready for AI. When AI is designed for them, AI -based pest surveillance, crop sap integration is our mantra. By integrating geospatial analytics with post -surveillance, we have delivered early warnings to cotton -growing farmers, reducing crop vulnerability and finance risk. This is predictive governance in action. Agriculture data exchange is also one thing which is defining this step. We are building a statewide interoperable agriculture data exchange. We are building a statewide interoperable agriculture data exchange. based on open standards and strong data governance. Data must empower farmers, not exploit them. Traceability digital public infrastructure in today’s global markets, the transparency is a mantra. We are unveiling a blueprint for a traceability DPI that will ensure end -to -end visibility across value chains, enhancing food safety, export competitiveness, and consumer trust.

And this is not proprietary. It is being designed as a replicable public infrastructure model for India and the entire global south. In partnership with India AI, by mission, the Government of Maharashtra the World Bank, and the Wadhani AI, we launched a global call for AI use cases in agriculture. The resulting compendium of real -world AI applications in agriculture was released in Delhi on 17 February 2026. This compendium documents successful AI deployments from Africa, Asia, Latin America, and beyond. India is convening global knowledge for the benefit of the global south. As we move towards AI for Agri 2026 in Mumbai, our vision rests on four pillars. Responsible governance. AI must be transparent, auditable, and explainable. Open and interoperable digital infrastructure.

innovation cannot scale in silos investment and scaling technology without capital remains just a theory and inclusion and gender equity is also a mantra 2026 is the international year of women in agriculture AI solutions must be designed with women farmers not merely for them Maharashtra today presents one of the most compelling agri -innovation ecosystems globally 150 lakh hectares of cultivated land diverse agro -climatic conditions leading agriculture universities and AI research centers a vibrant startup ecosystem a clear regulatory framework and single window facilities a vision for investors and a vision for the future We invite venture capital funds, impact investors, multilateral development banks, corporate innovation arms, and philanthropic foundations to partner with us. And in this partnership, we envisage scaling AI advisory platforms, co -developing traceability DPI modules, investing in agri -tech startups, supporting digital literacy, especially among women farmers, building capacity in the rural AI ecosystems.

When you invest in Maharashtra, you invest. In scalable solutions for engaging economies worldwide, food security, climate resilience, and AI governance are deeply connected. that master AI -enabled agriculture will secure farmer incomes and strategic stability. India has the scale, DPI, and democratic governance model to demonstrate how AI can be deployed responsibly at population scale. Maharashtra is proud to be laboratory of that ambition. Friends, this satellite session is a declaration. We will move from pilots to platforms, from fragmented data to interoperable systems, from experimentation to execution, from intention to investment. The government of Maharashtra stands ready to collaborate with the government of India, with states, with global institutions, investors, researchers, and farmer organizations. Let us ensure that AI becomes a force.

for food security, climate

Vikas Chandra Rastogi

Thank you, sir, for your visionary address. You always continue to inspire us to aim higher and achieve better. And under your leadership, I can assure you the Agriculture Department will rise to the challenge and serve the aspirations of more than 15 million farmers of the state of Maharashtra. Thank you so much, sir. We will now start the panel discussion. in a few moments. Thank you. Thank you. Thank you. Thank you. this session. We are fortunate to have with us a distinguished panel representing national policy leadership, global development, scientific expertise, national AI architecture, and digital public infrastructure innovation. Let me introduce the panelists once again. Dr. Devesh Chaturvedi, he is the Secretary, Ministry of Agriculture and Farmer Welfare.

Dr. Chaturvedi leads our national effort in agriculture and farmer’s welfare. Mr. Johannes Jett, he is the Regional Vice President, World Bank. Mr. Jett brings a vital global perspective on development and finance from the World Bank. Ms. Soumya Swaminathan, she is the Chairperson of Dr. M.S. Swaminathan Research Foundation. Dr. Swaminathan is a global leader in science, a champion for sustainable development, and a strong advocate for mainstreaming women farmers’ roles. in agriculture. Mr. Shankar Maruwala is a co -founder and CEO of Ekstey Foundation. He is a pioneer in building digital public infrastructure that empowers people at scale and I am very proud to say that the government of Maharashtra and Ekstey Foundation together have brought out Mahavistar which more than 2 .5 million farmers are using today to get the advisories and information that they need on a daily basis.

The objective of this panel discussion is to move from vision to implementation. Specifically, we will deliberate on how to institutionalize AI within agriculture systems at scale, how to ensure inclusion, especially of women farmers and smallholders, how to build interoperable, trustworthy and sustainable AI governance ecosystems, and how to strengthen collaboration between the center and the center. states, global institutions, industry, and academia. The session is also an important precursor to AI4Agree 2026 Global Conference where we will continue these deliberations in greater operational depth with governments, investors, innovators, and development partners. AI4Agree Conference is being held in Mumbai on 22nd and 23rd of February at Jio World Convention Center. With this context, let’s begin our discussion. My first question is to Dr. Devik Chaturvedi.

Sir, under your leadership, the ministry has taken significant steps in advancing the digital agriculture mission and operationalizing the Agri -Stack framework. You are laying a strong digital foundation for the sector. As we now look, at integrating AI more systematically into agriculture, how do you envision the central state collaboration framework, specifically to ensure that AI deployments are aligned with national architecture while allowing states the flexibility to innovate based on local agroclimatic and socioeconomic context? And finally, how can we institutionalize this collaboration to achieve population scale impact while mentoring interoperability and data trust?

Devesh Chaturvedi

Thank you. A lot of questions in the same question. So what I’ll do is I’ll just first take you through the initiatives. First of all, we deeply appreciate the leadership taken by Maharashtra under obviously the leadership of our honorable chief minister and with the agriculture department. They have done exceptional work in digital agriculture mission by developing farmer IDs and digital computers. We’ve done a lot of survey. and also they launched Mahavistar as a precursor of Bharatvistar. And recently on 17th, government of India have also launched one of the first integrated AI -based system for the farmers, which is Bharatvistar, which presently is undertaking providing services both through the app, Android -based app, as well as through mobile telephony on weather advisories, ICR -based crop advisories, pest advisories, market information regarding various agriculture produced, traded in the Mondays, and lastly, the government schemes of government of India.

Now, why is this important, AI is important in agriculture? Like we did a lot of, we started with digitalization of services, different services, we had DBT, we had online systems of applying for various, a common person applying to the common services, and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of and we started to have a lot of But what was felt was that while we had initiated this process to ensure that the bureaucratic red tapism is removed, what we were moving towards was a sort of digital red tapism.

Because within our ministry, different schemes had different apps. And they had different ways of selection. And within the state also, horticulture had a different database of farmers. Agriculture had a different database. Animal health has a different database. Crop insurance has a different database. So basically, a farmer who has to avail so many services, we felt that he or she was getting lost in which app to use for which. And sometimes it becomes more difficult to avail the services through online systems or to get advisories than to go to a person and say, tell me how to do it. So the whole idea was that once we have this AI -based system, we have a same platform for different… applications and different advisories at a click of the button or maybe just as a voice.

So that is the whole idea of shifting towards AI -based solutions. So now what we have initially in the first phase in the artificial intelligence system, the Bharat Vistar or the Mahavistar of Maharashtra, is that the advisories, the crop advisories, the weather advisories, schemes information, information about how to apply and what is the status of that application and also the Monday rates, all these have been put in the one platform. You can just make a presently it is working in English and Hindi but in next three to six months we’ll be taking it towards all the Bhashani related languages. And the next step is as you mentioned that the states are working together with us for the digital public infrastructure.

So close to 9 crore farmer IDs have been developed. So what is a farmer ID and you must have read the statement of Honourable Finance Minister that DPI is the new UPI. so what is the basic this agri stack which is the part of DPI is that for agriculture is that we have each farmer has a unique farmer ID with the back end all the crops the person has sown, what is the land available to that person, all the data with the share of the land and the crop sown and the soil health card details if the soil health has been given so with these basic details available on the system it empowers the farmer through that ID to avail services because it is already approved by the relevant authorities in the government so the person does not have to or the authorities who are giving the services are not required to cross verify the credentials of the farmer based on those those based on the record of rights or maybe the whatever it was in the different states so every state in Maharashtra is one of the leading states here we are working together to have a saturation of farmer IDs and crop survey and once this is there then this AI will further transform into a very very tailored advisor so a person calls or gives the farmer idea to Aadhaar and at the back end we will based on the consent access the details of where the farmer is from, what is the crop being grown, what is soil hand conditions and very targeted advice will be given which will be made operational in next 3 to 6 months so instead of pushing data which may not be of interest of the farmers very specific tailored data for that farmer will be available based on integration of digital public infrastructure with Bharat Vistar and the third aspect will come when we do the predictive models and we tried that and you must have remembered in the inaugural session when Google CEO mentioned about that predictive model which we did about 3 .8 crore farmers we used 100 years data of IMD and a model to predict a monsoon for the next 1 month and for next week and that prediction was fairly accurate and farmers, we got the feedback the farmers did take that decision to sow and to irrigate based on the predictive model which was sent.

And now we will expand the predictive models to ensure that we get more advisories of the market situation, of the weather situation, which will help improving the decision making of the farmers and so that they can increase their productivity, reduce their costs. So that is the whole idea of AI in agriculture. And we hope that more and more farmers will adopt it and it will be not exactly a replacement but a sort of additionality to the human, we can say extension services which we find is not able to reach to the farmers because of the resource constraints of each state. The extension machinery, the KVKs or our state extension machineries, it’s very difficult to reach each and every farmer because of the fact that we can’t have a person sitting in each village reaching to each farmer.

But AI along with digital public infrastructure, along along with the mobile and internet penetration in the various rural areas, will ensure that that gap is removed and we get more and more access to the farmers on

Vikas Chandra Rastogi

model that provide just -in -time support to central and state governments, enabling them to experiment, iterate, and scale AI solutions responsibly.

Johannes Zutt

Thanks very much for those questions, and thank you also for the invitation to be here today. So we’re on the cusp of a major revolution in how support to farmers and agriculture is happening. I actually grew up on a farm. I worked on a farm from the ages 10 to 21. I think every hour I wasn’t in school, that I was actually at home. I was working in a farm. In some ways, it feels paleolithic, because we didn’t have computers. We had telephones that were connected to wires, and our ability to get information about what was happening around us was extremely limited. We spent a lot of time trying to find out the things that today you can find out very, very quickly using small AI for agriculture.

And that’s truly right. evolutionarily empowering for farmers. But, you know, to make that work for farmers, there’s a lot of things that need to go right. And I think it’s worth reflecting a little bit on the different roles that different actors in the ecosystem have, starting obviously with government. My colleague mentioned a number of these things earlier. The government’s responsibility is principally on foundations, communications, things like the governance of AI, the interoperability, obviously ensuring that educational programs include appropriate types of skilling in the use of digital services. This is a big challenge in countries like India, where frankly there are still people who don’t have sufficient literacy to read what comes over a basic smartphone ensuring that the research and extension…

Thank you. that is provided through these small AI platforms is credible, is trustworthy, is backed by science. I think that’s also extremely important. Of course, farmers will find out if they aren’t, but at high expense, right? So we want to make sure that they’re not being advised to do things that are negative for them. And then also looking at the costs of service, the connectivity, what does the farmer actually need to be able to link into these different types of platforms that give information? Because, of course, we’re often also talking about farmers who have very, very few assets and who may be essentially unable to stay permanently connected or who are not able to stay permanently connected.

They’re not able to stay permanently connected or even easily connected to the Internet. They’re going to have very basic smartphones, et cetera. So the government has a lot of… of work to do in all of those areas. Then you can look at what can the private sector do. Now, one thing that the government needs to do is encourage crowded and private sector capacity and capital. But once we turn to the private sector, what is the private sector’s principal advantage? I think that there’s a lot of creativity in the private sector. So the actual applications that are being developed are being developed by individuals in the private sector with a passion for specific sorts of issues that are constraining farmer success.

And that creativity will result in a number of different applications that will be aimed, in most cases, to help farmers overcome certain hurdles that they face. And we can kind of let a thousand flowers bloom there. And see what actually takes root. And it’s amazing what you start to see. Just yesterday I was learning about an application in Morocco developed by a tomato farmer who was able to give advice about how much water tomato plants need simply by taking a picture of the current tomato plant. Take a picture and it tells you how much water you actually need to give this plant, which obviously in a water -stressed environment is vital, vital information. And then there are roles for institutions like my own, the World Bank Group, which can help to provide some of the financing that helps develop these applications and also the foundational backbone for artificial intelligence.

And we can also play a role at the advisory end where we are helping to truth test, if you like, the information that’s coming through different applications that are coming in. Coming out of the AI sandbox in different contexts to make sure that it’s… actually providing information that’s useful to the end beneficiary and enhancing from a productivity perspective at the farm level.

Vikas Chandra Rastogi

Thanks. I think you have rightly pointed out the role of innovation and research. And what we see is we require high -quality, robust data to actually build upon that. And as Honorable Chief Minister mentioned, Maha AgEx is one step in that direction wherein we bring diverse data sets and make them accessible to researchers, academic institutions, departments, and also startups. And many of these startups we will see they are showcasing their innovations in AI for Agri conference in Mumbai. So we’ll request all of you to please come there and see for themselves what kind of excitement they have and what kind of solutions our MDA says. I have one supplementary question to you. How do you see a platform such as…

AI Impact Summit as well as AI for Agree Global Conference, contributing to deeper global collaboration and south -south knowledge exchange in this domain?

Johannes Zutt

Thank you for that additional question. I mean, obviously, India is in a great position to lead the development of AI, particularly for developing countries where there are still significant challenges helping poor people to escape poverty permanently. India has demonstrated digital innovation for a long period of time already. It’s got an enormous population with a huge variety. The challenges of bringing farmer -appropriate data to the farmers’ fingertips in India are… I was going to say India is a microcosm of the rest of the world. It’s hardly a microcosm. It’s so huge. But because you have so many languages, so many different regions, so many different types, so many different cultures, and the starting conditions at the farm level are so incredibly varied, figuring out how to make AI at the farm level work in India will automatically have a large number of spillover learnings for other countries around the world.

And because India, after China and the United States, is the country in the world that is best positioned actually to push all of this work forward, and because it is itself a developing country, it’s very, very clear that it will have a central role to play in South -South learning for those reasons.

Vikas Chandra Rastogi

Thank you so much. I move on to Dr. Swaminathan. Dr. Swaminathan, your father, Professor M.S. Swaminathan, played a historic role in shaping India’s agriculture transformation during the Green Revolution, ensuring food security at a critical juncture in our history. Today, as we speak of a new phase of transformation driven by AI, we are again at an inflection point. You have consistently championed science -based policy, sustainability, and the empowerment of women farmers. With 2026 being recognized internationally as the year of women farmers, how can we ensure that AI -led agriculture transformation strengthens women’s agency, knowledge access, and climate resilience? And what institutional safeguards and design principles must we embed today so that this new technological revolution becomes equitable, farmer -centric, and grounded in scientific integrity?

Dr. Soumya Swaminathan

Thank you very much for that question, Vikasji. Not only is this year the International Year of the Woman Farmer, but we know that agriculture itself is increasingly being feminized, with many men actually leaving farming to the women and migrating out. To the cities for other opportunities. So it is really essential to put women… at the center of all that we are discussing. And I think the Chief Minister today gave us a wonderful vision of what can be the future, provided, of course, like you said, that there are the guardrails, there are the institutions, there are the safeguards and the design principles that we think about from the very beginning. So my father, Professor M.S. Parminathan, used to say that the Green Revolution was not only about the seeds.

Of course, the seeds played a very big role. You know, the high -yielding varieties. But it was about the entire ecosystem and the institutions that were developed at that time, which included the outreach, the, you know, later on the Krishi Vigyan Kendras, of course, were developed, but also the access to credit, the water, the fertilizers, the education and empowerment, and ultimately became a success because farmers realized the potential of it and took it on. So. And what he used to say is that, you know, every technology. No technology is pro -poor or pro -rich or pro -woman or against women. It’s how we use that technology. So it’s really, like you said, the inflection point today is how do we use this very powerful technology that’s come to us.

So I think there are a few points here to make sure particularly that women farmers are not left behind. The first important fact is that women in India, the minority of them who have their name on the land document, so mostly it is in the man’s name, and Deveshji was telling me today that this is improving and that the latest census shows that perhaps at least a quarter of the properties are also in the name of women, either jointly or – but that still means that, you know, three -fourths of them don’t have. And a system that operates – basically on publicly available data will then leave out those whose data sets are not available.

So I think – I think it would be really important at the early stages itself to think about how women’s data can be incorporated. Because the algorithms are fed by the data we have. And so all of these advisories may be very suitable for a man who’s operating a tractor on a farm, but not at all relevant for a woman who’s still working with outdated instruments and trying to, you know, till her land. And particularly when we look at more remote areas, tribal areas, where women do a lot of the agriculture, like millets, for example. Mostly it is women who grow millets. And there’s still a lot of mechanization which is absent completely. It is all still very much done using traditional methods and tools.

And it involves a lot of drudgery. So I would say that, you know, one of the benchmarks that I would look at is, is it reducing the drudgery and the workload on women farmers? Is AI helping to do that? So I think we also need to think about that. We also need to look at certain indicators for success. And you mentioned science. I mean, I’m a medical. researcher and the way that we evaluate products is by doing clinical trials, by examining the data and the evidence and then recommending it for wider use. So again, a note of caution would be to, as we roll it out, we need innovation certainly. We also need to do the evaluation, looking at inherent biases, looking at who’s being excluded, looking at are there unanticipated risks or side effects that we didn’t know about.

But most of all, it’s this inclusion. I think we don’t want those who are already left behind to be further left out. So I think the ongoing research and data collection and feedback loops and most importantly, having the voices of those for whom we are developing all these. I think in the room, I don’t think we have any farmers or women farmers. So we are all discussing from what we know. But if you’re the farmer, like you were saying, working there and you know the constraints and the which you’re working. So I think the women farmers and farmers in general must have a role. They must be part of these committees that evaluate or make recommendations or make suggestions on improvement.

It has to be an iterative process. I think any technology is as good as the application for which it’s developed. I’ll give you one example of an app that the MS Farminathan Research Foundation developed for fisher women. We had a very successful app for fishermen called the Fisher Friendly Mobile App that won the UN Tech for Nature Award last year. But fisher women were as usual left out. And so the Women Connect app actually gives them on a tablet information that they need to sell. Because once the fishermen have come back from seeds, the women who have to do all of the post -harvest, and the same is true for crops or fruits or vegetables as well.

So that connection to the market, of course the information about pests and pathogens and when to buy what and what inputs to use. But also being able to organize themselves. And I think women And there are many FPOs now and FPCs and SHGs made of women farmers, empowering them and giving them the knowledge and tools. And the last thing I would say is we still need humans in the loop. I don’t think we should think that completely making everything run by machines is going to solve our problems. I think it’s risky there. And in a country like India, we also need employment. And so we should think of, and I don’t know how many of you have seen this film called Humans in the Loop.

But it’s a tribal woman from Jharkhand who actually raises questions about the algorithm. It’s a very thought -provoking film. So I think Humans in the Loop is going to be important. We have our Krishis, Sakhis and so on. We need to empower them with these. So I think AI and all these digital tools, if they’re used in addition to the traditional knowledge and wisdom that people have and augment it and give them at the right time, at the right place, the knowledge they need, I think we can go a very long way. Thank you.

Vikas Chandra Rastogi

Thank you, madam. You have rightly. pointed out the need to be more sensitive while developing systems for inclusivity and to ensure that for whom they are being developed and they are in the loop and they are being consulted. In fact the feedback mechanism that we have developed in Mahavistar takes care of those requirements. I am also very happy to share that Government of Maharashtra and Dr. M.S. Swaminathan and his foundation are working together on some of these issues in terms of how to bring women’s right in farming at the center stage how do we create bio -happiness using our universities and educational systems and what kind of nutritional security we must look for because we have food security but it’s the nutritional security that we must aspire for.

We are happy to have support and assistance from MSSRF. in that direction. My final question is to Mr. Shankar Marubala. Mr. Shankar, XTAP has played a foundational role in shaping India’s DPI landscape through open source platforms such as Sunbird, which has powered large scale systems like Diksha, Mahavistar, and open network initiative built on backend protocol. These efforts have demonstrated how open standards and interoperable architecture can enable population scale transformation that we are already seeing today. As we now enter the era of AI driven public systems, how should we think about standardizing AI based ecosystem in a similar spirit? How can we bring DPI into AI? And what architecture and governance principles are required to ensure interoperability, trust and sustainability in AI deployments across sectors such as agriculture?

Shankar Maruwada

Again, a whole lot of questions, but let me. I’ll make my best attempt to answer those. more than 100 years ago the world faced what was known as a malthusian crisis where malthus the economist predicted that if we continue to grow in the same way we’ll run out of land we’ll run out of soil we were a billion and a half then we are eight billion most of us may not even have heard of the malthusian crisis what happened someone called haber and someone called bosch created a miracle haber synthesized ammonia using high pressure and temperature and bosch put it into an industrial process that phenomena is now historically known as pulling bread out of air it took a lot of effort and as samya said creation of a massive ecosystem germany which pioneered this lost that race to us because us did a better job of diffusing the technology safely to the farmers.

They created the discipline of agriculture engineering. They created institutions like the Fertilizer Development Center. They held technology demonstrations to farmers to show them how synthetic ammonia could be used. By the way, 50 % of the nitrogen in our body comes from synthetic nitrate ammonia. That’s a fact. We owe a lot to Heber and Bosch. China then took it on in the 80s by buying 10 big plants from Kellogg, training 300 million farmers, showing them how to use synthetic fertilizers. They went on to be the global leaders in agriculture. India is at a point where if you learn the lessons from such past things, our green revolution, our DPI experience, we are at a pivotal point where the equivalent of pulling bread out of thin air is pulling intelligence from the earth and providing it to the farmer this is again not science fiction Mahavistar, the pioneer along with Bharatvistar have taken the first steps to this so when a Mahavistar was designed to build off what Swami has said, it was designed with inclusion in mind inclusion, diversity was not an afterthought because to solve for not just Maharashtra’s problems, for India’s scale and diversity, we need to think of the last person the most discriminated in the remotest part of India and design systems that work for them we call that DPI now let me give you a specific example of this in Bharatvistar right from the beginning the design specs was we need an illiterate farmer to build off John’s point about digital literacy with a feature phone not a smart phone to be able to talk in his or her native language and native dialect Marathi itself has many dialects right talk on the phone like the way she is comfortable talking to another person ask the question have a conversation get a bunch of answers that process took us the better part of nine months why because it’s not just AI it’s data it’s processes it’s training the farm extension workers it is having trust on will this work what about the costing will I blow up my entire stage budget on a model right do I have autonomy can I switch models out in and out these are very very difficult questions it took us in partnership with a whole lot of people and we are working on a I mean, Government of Maharashtra led the effort, but IndiAI Mission, Bhashini, IIT Madras, IIIT Hyderabad, World Bank, Google, many other providers, everybody chipped in the little part of the solution.

Now, here’s the best part. Because we all collaboratively invested in figuring out a solution there, that solution could be deployed in Bharat Vistar with more confidence easily. Again, the same challenges that Secretary Chaturvedi talked about, do we have the data? He used a very nice phrase, digital red tapism, right? Our data is in different formats. What matters is the intent of the government. The government of India, which triggered the process, which allowed Bharat Vistar to be launched the day before, it’s a start. Data will get better, the systems will get better, usage will improve, that will generate more data, and then over time, years, the ecosystem will be built. This we know from our experience.

What makes this happen? What is that secret sauce, the design principles? It is the same as DPI. What worked for DPI, we are taking those same principles. One, open interoperable systems. Think networks and not just portals and platforms and siloed and fragmented systems. What’s the best example of this? The railways in India. We have such a vast landscape, but the rails are common. Every state can decide what it wants to move, private, public, defense, farming. The Indian railways is just providing a backbone. That allows. Everyone to. . . . . . . . . do this. There was a time when we had different rail gauges. Right? Now, that sounds so silly, but there was a time like that.

India is showing that we don’t have to repeat those early mistakes in digital also. By creating interoperable networks based on open protocols like Beacon, by collaborating with each other, one of us is bringing in data, somebody is bringing in technology, somebody is bringing in policy, somebody is bringing in research. These collaborative open networks and with the launch of Bharat Vistar puts India in a very unique and responsible position. Unique because we have these open rails. We have the experience of DPI. Responsible because it is a start. Unlike the technologies of the past where you perfect the technology and then deploy AI. you deploy something minimum to start and then evolution models get better, data gets better, usage gets better and then it gets better and better over time.

That is the unique junction we are in in India. What will that mean? When ICAR plugs into this network with its weather and pricing data, that network makes it available to any state that wishes to turn on the supply from ICAR. When a private sector comes out with a very innovative app, let’s say the tomato example that John talked about, any state can say, I like that. I think I will have that made available to my farmers. For the farmers, they anyway trust the state. They can go to the same app and now see this also there. If the tomato app person wants, they can go. They can go directly to each farmer. very expensive.

So Shared Rails allows us to spread innovation, diffuse it very quickly through society, keeping in mind both inclusion and rewarding innovation because innovation has to be rewarded. And I want to end with a very simple analogy. When Edmund Hillary climbed Mount Everest, he made a lot of people believe it is possible. When Mahavistar was launched, it made the country believe that it is possible to make AI serve the farmer. And to that extent, the responsibility that Mahavistar, Maharashtra government and government of India has is to create these pathways for the rest of the country for the other states. At XTEP Foundation, we made a declaration two days ago. We would like to see a world by 2030 where there are hundreds, hundreds such diffusion power.

pathways each created by a different set of people in different sectors in different countries and continents but each inspiring different AI pathways to safe impact at scale and it’s a very exciting vision it’s a very collaborative vision if you all get together we can also create miracles in our own lifetime thank you

Vikas Chandra Rastogi

with that profound thought we’ll conclude today’s panel discussion I thank all the panelists they have really opened a new vision in front of all of us and we’ll invite all of you to AI for Agree conference in Mumbai on 22nd thank you so much we don’t have question actually a time to question the next session is about to start we can discuss that Thank you. Thank you.

Devendra Fadnavis

Speech speed

92 words per minute

Speech length

Inclusion, Gender Equity & Human‑in‑the‑Loop

Topics

Closing all digital divides | Capacity development | Social and economic development

Johannes Zutt

Speech speed

146 words per minute

Speech length

Inclusion, Gender Equity & Human‑in‑the‑Loop

Topics

Closing all digital divides | Capacity development | Information and communication technologies for development

Devesh Chaturvedi

Speech speed

174 words per minute

Speech length

AI Applications for Farmer Empowerment & Climate Resilience

Topics

Artificial intelligence | Social and economic development | Closing all digital divides

Shankar Maruwada

Speech speed

134 words per minute

Speech length

Inclusion, Gender Equity & Human‑in‑the‑Loop

Topics

Closing all digital divides | Capacity development | Information and communication technologies for development

Dr. Soumya Swaminathan

Speech speed

176 words per minute

Speech length

1140 words

Speech time

387 seconds

Women Farmers at the Centre of AI Design

Explanation

Swaminathan stresses that AI tools must be built around women’s land‑rights, labor realities and should be evaluated for bias, ensuring that women are not left behind in digital agriculture.

Evidence

“So I think it would be really important at the early stages itself to think about how women’s data can be incorporated” [112]. “It is really essential to put women… at the center of all that we are discussing” [114]. “We also need to do the evaluation, looking at inherent biases, looking at who’s being excluded, looking at are there unanticipated risks or side effects that we didn’t know about” [118].

Major discussion point

Inclusion, Gender Equity & Human‑in‑the‑Loop

Topics

Closing all digital divides | Human rights and the ethical dimensions of the information society | Capacity development

Iterative, Evidence‑Based Evaluation of AI Tools

Explanation

She calls for clinical‑trial‑style testing of AI applications, continuous feedback loops, and inclusion of farmer voices to ensure tools reduce drudgery and deliver real benefits.

Evidence

“It has to be an iterative process” [115]. “One of the benchmarks that I would look at is, is it reducing the drudgery and the workload on women farmers?” [116]. “researcher … by doing clinical trials, by examining the data and the evidence and then recommending it for wider use” [120]. “the ongoing research and data collection and feedback loops and most importantly, having the voices of those for whom we are developing all these” [121].

Major discussion point

Inclusion, Gender Equity & Human‑in‑the‑Loop

Topics

Monitoring and measurement | Human rights and the ethical dimensions of the information society | Capacity development

Vikas Chandra Rastogi

Speech speed

102 words per minute

Speech length

Inclusion, Gender Equity & Human‑in‑the‑Loop

Topics

Closing all digital divides | Capacity development | Social and economic development

Agreements

Artificial intelligence | Capacity development | The enabling environment for digital development

Similar viewpoints

Government officials share a unified vision of scaling AI platforms from pilots to population-scale implementation, with Maharashtra’s Mahavistar serving as a model for national expansion through Bharatvistar

Speakers

– Vikas Chandra Rastogi
– Devendra Fadnavis
– Devesh Chaturvedi

Arguments

Maharashtra has launched Maha Agri AI Policy 2025-2029 with platforms like Mahavistar serving 2.5 million farmers and supporting multilingual advisories including tribal languages

AI must be built on trusted data, ethical governance, and public accountability, moving from pilots to platforms at population scale

Government of India launched Bharatvistar as an integrated AI-based system providing weather advisories, crop advisories, pest alerts, market information, and government schemes access

Topics

Artificial intelligence | Information and communication technologies for development | Social and economic development

Both emphasize that inclusive design must be intentional from the beginning, with continuous feedback from end users, particularly focusing on reducing burden for marginalized farmers

Speakers

– Dr. Soumya Swaminathan
– Shankar Maruwada

Arguments

AI should reduce drudgery and workload on women farmers, with evaluation processes including farmer voices and feedback mechanisms

AI systems must be designed for inclusion from the beginning, solving for India’s scale and diversity by considering the most discriminated in remote areas

Topics

Closing all digital divides | Human rights and the ethical dimensions of the information society | Artificial intelligence

Both see India’s diversity as an advantage for developing AI solutions that can be applied globally, emphasizing the importance of collaborative approaches for successful deployment

Speakers

– Johannes Zutt
– Shankar Maruwada

Arguments

India’s experience with diverse languages, regions, and farming conditions positions it to lead AI development for developing countries with significant spillover learnings

The collaborative approach involving multiple stakeholders allows solutions to be deployed with confidence across different contexts

Topics

Artificial intelligence | Information and communication technologies for development | Capacity development

Unexpected consensus

Emphasis on human oversight and gradual implementation rather than complete automation

Speakers

– Dr. Soumya Swaminathan
– Shankar Maruwada
– Devesh Chaturvedi

Arguments

Technology evaluation requires ongoing research, data collection, feedback loops, and keeping humans in the loop rather than complete automation

Open interoperable systems based on DPI principles are essential, with collaborative networks allowing states to share innovations while maintaining local autonomy

Government of India launched Bharatvistar as an integrated AI-based system providing weather advisories, crop advisories, pest alerts, market information, and government schemes access

Explanation

Despite the focus on AI advancement, there is unexpected consensus on maintaining human oversight and gradual, iterative implementation rather than pursuing complete automation, showing a cautious and responsible approach to AI deployment

Topics

Artificial intelligence | Human rights and the ethical dimensions of the information society | Monitoring and measurement

Strong focus on women farmers and gender equity across all speakers

Speakers

– Devendra Fadnavis
– Dr. Soumya Swaminathan
– Vikas Chandra Rastogi

Arguments

2026 being the International Year of Women Farmers requires AI solutions designed with women farmers, not merely for them, addressing the feminization of agriculture

Women farmers often lack land ownership documentation, risking exclusion from AI systems that rely on publicly available data, requiring specific design considerations

Maharashtra has launched Maha Agri AI Policy 2025-2029 with platforms like Mahavistar serving 2.5 million farmers and supporting multilingual advisories including tribal languages

Explanation

There is unexpected strong consensus across government officials and researchers on prioritizing women farmers and gender equity, which goes beyond typical technology discussions to address fundamental social equity issues

Topics

Closing all digital divides | Human rights and the ethical dimensions of the information society | Artificial intelligence

Overall assessment

Summary

The speakers demonstrate remarkable consensus on key principles: the need for open, interoperable AI systems built on ethical foundations; the critical importance of inclusive design that prioritizes marginalized farmers, especially women; the necessity of comprehensive ecosystem approaches involving multiple stakeholders; and the value of gradual, human-supervised implementation over complete automation. There is also strong agreement on India’s potential to lead global AI development for agriculture while maintaining responsible governance.

Consensus level

Very high level of consensus with no significant disagreements identified. This strong alignment suggests a mature understanding of both the opportunities and challenges in AI deployment for agriculture, with implications for successful policy implementation and international collaboration. The consensus particularly around inclusion and ethical governance indicates a responsible approach that could serve as a model for other developing countries.

Differences

Different viewpoints

Approach to AI deployment and evaluation

Speakers

– Dr. Soumya Swaminathan
– Shankar Maruwada

Arguments

Technology evaluation requires ongoing research, data collection, feedback loops, and keeping humans in the loop rather than complete automation

The collaborative approach involving multiple stakeholders allows solutions to be deployed with confidence across different contexts

Summary

Dr. Swaminathan advocates for a cautious, medical research-style evaluation approach with extensive testing before deployment, while Shankar Maruwada supports a ‘deploy and iterate’ model where minimum viable solutions are launched and improved over time

Topics

Artificial intelligence | Human rights and the ethical dimensions of the information society | Monitoring and measurement

Role of human oversight in AI systems

Speakers

– Dr. Soumya Swaminathan
– Devesh Chaturvedi

Arguments

We still need humans in the loop. I don’t think we should think that completely making everything run by machines is going to solve our problems

AI along with digital public infrastructure, along along with the mobile and internet penetration in the various rural areas, will ensure that that gap is removed and we get more and more access to the farmers

Summary

Dr. Swaminathan emphasizes the critical need for maintaining human oversight and warns against complete automation, while Secretary Chaturvedi focuses more on AI’s ability to bridge gaps in extension services without emphasizing human oversight

Topics

Artificial intelligence | Human rights and the ethical dimensions of the information society | Capacity development

Unexpected differences

Emphasis on data governance versus service delivery

Speakers

– Devesh Chaturvedi
– Devendra Fadnavis

Arguments

Digital agriculture mission has developed 9 crore farmer IDs creating a foundation for AI integration, with Maharashtra leading in farmer ID saturation and crop surveys

AI must be built on trusted data, ethical governance, and public accountability, moving from pilots to platforms at population scale

Explanation

While both are government officials working on the same initiative, the Secretary focuses heavily on the technical infrastructure and data collection aspects, while the Chief Minister emphasizes governance principles and ethical considerations. This represents different priorities within the same government framework

Topics

Data governance | Artificial intelligence | The enabling environment for digital development

Overall assessment

Summary

The discussion shows relatively low levels of direct disagreement, with most differences emerging around implementation approaches rather than fundamental goals. Key areas of difference include the pace and methodology of AI deployment (cautious evaluation vs. iterative deployment), the role of human oversight, and the balance between technical infrastructure development and governance principles.

Disagreement level

Low to moderate disagreement level with significant implications for AI governance in agriculture. The differences in approach between cautious evaluation and rapid iteration could affect the speed and safety of AI deployment at scale. The varying emphasis on human oversight versus automation has important implications for employment and farmer agency. These disagreements, while not confrontational, represent different philosophies that could lead to different policy outcomes and implementation strategies.

Partial agreements

Both speakers agree on the critical importance of inclusive AI design, but disagree on implementation approach – Dr. Swaminathan emphasizes post-deployment evaluation and farmer participation in committees, while Shankar advocates for inclusive design from the beginning as a foundational principle

Speakers

– Dr. Soumya Swaminathan
– Shankar Maruwada

Arguments

AI should reduce drudgery and workload on women farmers, with evaluation processes including farmer voices and feedback mechanisms

AI systems must be designed for inclusion from the beginning, solving for India’s scale and diversity by considering the most discriminated in remote areas

Topics

Closing all digital divides | Human rights and the ethical dimensions of the information society | Artificial intelligence

Both speakers recognize the need to address women farmers’ specific needs, but approach it differently – the Chief Minister focuses on participatory design principles, while Dr. Swaminathan identifies specific structural barriers like land ownership documentation that could exclude women from AI systems

Speakers

– Devendra Fadnavis
– Dr. Soumya Swaminathan

Arguments

AI solutions must be designed with women farmers not merely for them, addressing the feminization of agriculture

Women farmers often lack land ownership documentation, risking exclusion from AI systems that rely on publicly available data, requiring specific design considerations

Topics

Closing all digital divides | Human rights and the ethical dimensions of the information society | Artificial intelligence

Similar viewpoints

Government officials share a unified vision of scaling AI platforms from pilots to population-scale implementation, with Maharashtra’s Mahavistar serving as a model for national expansion through Bharatvistar

Speakers

– Vikas Chandra Rastogi
– Devendra Fadnavis
– Devesh Chaturvedi

Arguments

Maharashtra has launched Maha Agri AI Policy 2025-2029 with platforms like Mahavistar serving 2.5 million farmers and supporting multilingual advisories including tribal languages

AI must be built on trusted data, ethical governance, and public accountability, moving from pilots to platforms at population scale

Government of India launched Bharatvistar as an integrated AI-based system providing weather advisories, crop advisories, pest alerts, market information, and government schemes access

Topics

Artificial intelligence | Information and communication technologies for development | Social and economic development

Both emphasize that inclusive design must be intentional from the beginning, with continuous feedback from end users, particularly focusing on reducing burden for marginalized farmers

Speakers

– Dr. Soumya Swaminathan
– Shankar Maruwada

Arguments

AI should reduce drudgery and workload on women farmers, with evaluation processes including farmer voices and feedback mechanisms

AI systems must be designed for inclusion from the beginning, solving for India’s scale and diversity by considering the most discriminated in remote areas

Topics

Closing all digital divides | Human rights and the ethical dimensions of the information society | Artificial intelligence

Both see India’s diversity as an advantage for developing AI solutions that can be applied globally, emphasizing the importance of collaborative approaches for successful deployment

Speakers

– Johannes Zutt
– Shankar Maruwada

Arguments

India’s experience with diverse languages, regions, and farming conditions positions it to lead AI development for developing countries with significant spillover learnings

The collaborative approach involving multiple stakeholders allows solutions to be deployed with confidence across different contexts

Topics

Artificial intelligence | Information and communication technologies for development | Capacity development

Takeaways

Key takeaways

AI in agriculture must move from pilot projects to population-scale implementation with Maharashtra’s Mahavistar serving 2.5 million farmers as a successful model

Digital Public Infrastructure (DPI) principles of openness and interoperability are essential for AI agriculture systems, similar to how railway networks operate with common rails but diverse applications

Central-state collaboration framework should leverage the 9 crore farmer IDs already developed while allowing states flexibility to innovate based on local contexts

Women farmers must be central to AI agriculture design, especially given the feminization of agriculture and 2026 being the International Year of Women Farmers

AI systems require trusted data, ethical governance, and public accountability to achieve scale, moving away from fragmented digital systems that create ‘digital red tapism’

India’s diversity in languages, regions, and farming conditions positions it uniquely to lead AI development for the global south with significant spillover benefits

Technology evaluation must include ongoing research, farmer feedback loops, and keeping humans in the loop rather than complete automation

Collaborative ecosystems involving government, private sector, research institutions, and international organizations are essential for successful AI deployment

Resolutions and action items

AI for Agri 2026 Global Conference scheduled for February 22-23 in Mumbai at Jio World Convention Center to continue operational discussions

Government of Maharashtra to collaborate with MS Swaminathan Research Foundation on women’s rights in farming, bio-happiness creation, and nutritional security

Bharatvistar to expand to all Bhashini-related languages within 3-6 months from current English and Hindi support

Integration of farmer IDs with AI advisory systems to provide tailored advice based on individual farmer data within 3-6 months

XTEP Foundation declared goal to create hundreds of AI diffusion pathways by 2030 across different sectors and countries

Maharashtra government inviting venture capital funds, impact investors, multilateral development banks, and philanthropic foundations for partnerships

Expansion of predictive models beyond weather to include market situations and other decision-making support for farmers

Unresolved issues

How to effectively include women farmers who lack land ownership documentation in AI systems that rely on publicly available data

Addressing connectivity and cost barriers for farmers with basic smartphones and limited internet access

Ensuring scientific credibility and avoiding negative advice in AI systems while maintaining innovation pace

Balancing employment needs with automation in a country requiring job creation

Standardizing evaluation processes and metrics for AI agriculture applications across different contexts

Managing the transition from fragmented departmental databases to integrated systems without disrupting existing services

Determining optimal governance structures for AI systems that balance innovation with safety and accountability

Suggested compromises

Using AI as ‘additionality’ to human extension services rather than complete replacement to address both scale needs and employment concerns

Implementing minimum viable AI systems that evolve over time rather than waiting for perfect technology before deployment

Creating shared digital rails (infrastructure) while allowing diverse applications and innovations on top

This powerful analogy reframes Mahavistar not just as a technological achievement but as a proof of concept that changes what’s perceived as possible. It elevates the discussion from implementation details to transformational vision and psychological impact.

Impact

This metaphor provided an inspiring conclusion that synthesized the entire discussion around the theme of demonstrating possibility rather than just delivering services. It shifted the final tone from technical details to visionary leadership, emphasizing Maharashtra’s role in creating pathways for others to follow.

Overall assessment

Explanation

This explores the ‘humans in the loop’ approach to ensure AI complements rather than replaces human agricultural workers.

Disclaimer: This is not an official session record. DiploAI generates these resources from audiovisual recordings, and they are presented as-is, including potential errors. Due to logistical challenges, such as discrepancies in audio/video or transcripts, names may be misspelled. We strive for accuracy to the best of our ability.